The range of information that can be searched using we-based search engines has been extended beyond text-based web documents to multimedia such as audio, image, and video.
Recently, interest in video search has increased along with widespread introduction of services that allow users to upload various videos to a site to share the various video with other users. A number of portal sites have launched search engines that can easily search for multimedia such as video.
Since most search engines perform search by analyzing and using file formats, file names, and metadata of multimedia as search elements, the engines only provide search results obtained based on such file formats, file names, and metadata that are determined when multimedia data is generated.
A conventional search engine, which converts voice included in multimedia content into caption data using a voice recognition technology and performs search using the caption data as a search medium, has also been developed. When providing multimedia content items including an input search term as search results, this conventional search engine primarily provides a list of titles of the multimedia content. Therefore, in order to find sections of the multimedia content items including the search term, it is necessary to play and check each section included in the multimedia content items one by one, thereby increasing inconvenience.
In addition, when converting voice into caption data using a voice recognition technology, the conventional search engine cannot identify sentences in the caption data. That is, the caption data produced through conversion of voice only includes a series of words and does not include periods such that it is not possible to discriminate between sentences in the caption data. Since it is not possible to discriminate between sentences including the search term, it is not possible to provide a solution that can search for media sections associated with sentences. This is because it is not possible to divide a media content item into a plurality of media sections that are synchronized with sentences, phrases, clauses, or words although it is possible to divide media content on a time or capacity basis to generate a plurality of media sections.
As a result, in the conventional technology, it is not possible to extract only sections (media sections) including the search term from all sections (media sections) of media content items found as search results such that it is not possible to stream only the media sections including the search term among all media sections of the media content items. Thus, there is a need to provide a novel solution that can search for only media sections that are synchronized with sentences, phrases, clauses, or words and can stream only the media sections as search results.